[XMLBEANS-311] XmlCursor.getTextValue returns concatenation of child text, comment, and processing instruction nodes

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: Version 2.2.1
• Fix Version/s: Version 2.2.1
• Component/s: Cursor
• Environment: Microsoft Windows XP Professional Version 2002 (Service Pack 2); Intel Pentium 4 CPU 3.00GHZ
• Assignee: Cezar Cristian Andrei
• Reporter: Coram Bryant
• Created: 31/Jan/07 23:36
• Updated: 01/Feb/07 00:06
• Resolved: 01/Feb/07 00:06

Description

According to the XmlBeans javadocs, if the XmlCursor corresponds to an element then non-text content such as comments and processing instructions are ignored. However, as the following test shows, this is not the case:

String xml = "<element><!-A Comment->999<?PI A Processing Instruction?></element>";
XmlObject xo = XmlObject.Factory.parse(xml);
XmlCursor xc = xo.newCursor();
xc.toFirstChild();
System.out.println(xc.currentTokenType()); // Verify that we are positioned at the start of an element
System.out.println(xc.getTextValue());

This prints out:
START
A Comment 999A Processing Instruction

Analysis:
In this case, the call stack is as follows:
org.apache.xmlbeans.impl.store.Cursor.getTextValue
org.apache.xmlbeans.impl.store.Cursor._getTextValue
org.apache.xmlbeans.impl.store.Locale.getTextValue

In this method, the text value is built up with the text of instances of Cur for which the isText method returns true. I believe the org.apache.xmlbeans.impl.store.Cur.isText method should be as follows:

boolean isText ( )

{ assert isPositioned(); return _pos > 0 && kindIsContainer(_xobj.kind()); }

This would exclude comments and processing instructions.